Monday, April 20, 2015

...

Why out of so many Boiler controls, steam Temperature control is so critical???


Functional aspects of superheat temperature control types in utility boiler. by: Swapan Basu*


*Author:  Power plant Instrumentation and control handbook Elsevier (http://store.elsevier.com/Power-Plant-Instrumentation-and-Control-Handbook/Swapan-Basu/isbn-9780128011737/ )

*Systems & Controls (Consulting Engineers I&C) Kolkata India. (basu.swapan@gmail.com)

 Hope you have enoyed the past week nicely. I enjoyed bengali new year "SHUBH NABA BARSHO" " শুভ  নববর্ষ " 

Now let us go back to the discussions. Earlier we had discussed PID controllers and DEB controls. This time our discussions will be on state variable controllers and mathematical model.

http://www.sciencedirect.com/science/book/9780128009406

5.0    State variable controller with observer (SCO)

In control engineering, a system can be described in terms of state space representation which is a mathematical model of physical system set by input output and state variables related by first order differential equation. So, state of a system is described by a set of variables; such that knowledge of these variables and input functions related by a set of differential equation (and  will describe dynamics and future state of  the system) will describe it. If X1,X2… Xn are state variables and  A= State matrix B=Input matrix, C= output matrix and D= transmission matrix; the system can be represented by X=Ax + BU (State differential Equation)  & Y= Cx +DU (output).
From the above, it is clear that the main feature of state variable controller it not only use of output /input variables but also use intermediate process conditions (which may not be possible to measure physically) to offer suitable dynamics (high gain factor without oscillation). A typical State variable control system (comparing with PID) has been presented in Fig. 2. To get the best results, state controllers are used with observers. In conventional state controller, adaptation of internal parameters to the load is not possible practically [6], so  there will be a permanent deviation in steady state. In order to take care of dead time and other special features, state controllers are normally provided with observers[6] such as


v  State Observer
v  Disturbance Observer
v  Dead time Compensator etc.

Lets now examine Fig . 3 : State variable controler.









This loop is characterized by high and load dependent time constants non linearity (higher order) system.  Standard step change is used to get the knowledge about the various parameters such as Gain , system order (3,5 etc.), time constants, dead time etc. with the help of special algorithms & observer. Function generators use gain, lag, dead time compensator etc. to develop X1,X2… Xn state variables (- - the current operating point as a function of load).  A typical State variable controller with observer has been shown in Fig 4 [6].  The state controller here has been adapted from Mauell, and has three observers (viz. disturbance, state & actuator) and one state controller. Factor v is speed factor, represent the factor by which the feedback loop to be faster than the controlled system. With SCO it is possible to reduce SH outlet temperature deviation by 30-40% and gain efficiency.












6.0    Mathematical model based apprach: Predictive adaptive  & dynamic control

As discussed earlier, major problem associated with this loop is the time variable process dynamics, variable time delay and sudden unpredictable events (e.g. dirt deposit).  Controllers based on mathematical model could be a solution to these. In this method based on process characteristics a transfer function is identified i.e. process disturbances are modelled to incorporate adaptive feed forward, in to the control strategy in the form of updating of control model as shown in fig .5A.  Based on this update & set point,   the controller output is calculated and applied to the process. Based on the process response transfer function as well as controller update model is readjusted to ensure that  desired output is achieved. Since the model uses mathematical model duly adjusted with process transients,  to predict the process response there will be less chances of overshoot.











These systems with expert systems has been shown in Fig 5B. The basic philosophy behind these are [7] :
·   Anticipating process using mathematical model as discussed
·   Adjusting model parameters  using adaptive system to make predicition error to zero.
·   Incorporation of process expert in to the system

Expert determines the domain of control. During predictive mode control block generates the predictive output, as per desired output from the drive block, to be applied to the process.  Based on process output adaptive system adjust the  control block parameter so that predictve output error is zero. During expert control, expert knowledge is applied to the systems to meet the untoward situations.  Dynamic modelling techonlogy(DMT) a part of Object modelling technique are also used in predicting correct modelling & get proper transient response of the system. Laguerre functions are better suitated for process transfer functions which are transient in nature. DMT produces a set of weights for Laguerre functions in a series such that (alike fourier transform where periodic signals are approximated by use of cosine function) when these are summed up a reasonable approximation of the original transient signal is obtained[3]. This control can be interfaced with DCS as OPC server and in case of failure of the same DCS can fall back to PID.  In case of supercritical (SC) and ultrasupercritical (USC) boilers FR/FW ratio are used as coarse control and spray as fine control. However the feedback signal of FR/FW cannot be fast & accurate at the same time. There will be delay while FR/FW changing and its response in MS temeprature. Therefore intermediate temperatures are necessary as in SCO. Even precise control of FR/FW control can give good result in steady state but there will be many other affecting factors.  Active disturbance rejection control (ADRC) is another model based approach to get  better result for uncertain non linear disturbances[8]. ADRC consists of Tracking differenetaitor, non linear combination and extended state observer[8]. This is well suited for two stage sprays mainly found in SC/USC boilers. Typical ADRC approach for secondary spray control have been shown in fig.6with other details in the associated box.








we meet again next week.
 “stay tuned for a new post next week…” 

Best wishes for you to enjoy the week and join me with fresh mind for discussions on Fuzzy things!!!!









No comments:

Post a Comment